Targeting Antioxidant Systems to Treat Pancreatic Cancer

Transcription

1 Targeting Antioxidant Systems to Treat Pancreatic Cancer by Francesco Gabriele Tatangelo Fazzari A thesis submitted in conformity with the requirements for the degree of Master of Science Department of Laboratory Medicine & Pathobiology University of Toronto Copyright by Francesco Gabriele Tatangelo Fazzari 2018

2 Targeting Antioxidant Systems to Treat Pancreatic Cancer Francesco Fazzari Master of Science Department of Laboratory Medicine & Pathobiology University of Toronto Abstract 2018 Pancreatic cancer has the worst prognosis of all cancers. Pancreatic tumours are characterized by excessive oxidative stress that may provide a unique vulnerability to target in the development of new therapeutic strategies. I investigated a novel combination therapy targeting the two main intracellular antioxidant systems, glutathione and thioredoxin. Buthionine sulfoximine (BSO) targets glutathione synthesis and auranofin targets thioredoxin recycling. These drugs have been tested in human clinical trials and are strong candidates for repurposing in cancer therapy. I found that BSO potentiated auranofin s cytotoxic effects in vitro. Furthermore, the combination augmented oxidative stress in cells. Auranofin engaged its target TrxR in vitro at concentrations causing cell death. I also assessed the biodistribution of auranofin in vivo and observed a preferential accumulation in the stroma 24 hours after treatment. Overall, I conclude that this treatment strategy can be effective against pancreatic cancer and deserves attention for future development. ii

3 Acknowledgments First and foremost, I would like to thank my supervisor, Dr. David Hedley, for providing an excellent opportunity to research in his lab. His guidance was instrumental in helping me to develop a successful project and for supporting me as I take the next steps in my career. I would also like the thank all the members of the Hedley lab past and present for their assistance throughout my tenure. They provided an enjoyable work environment and were always willing to lend a helping hand to teach me different techniques. Next, I would like to thank my committee members, Drs. Richard Hill and Marianne Koritzisnky. Their critical assessment of my work helped to shape my project to what it has become. Lastly, I would like to thank my family and friends for providing support throughout my studies, and for helping me to enjoy my time outside of the lab. This work was made possible by support through funding from the Ontario Centres of Excellence. My stipend was partially supported by the University of Toronto Fellowship Award. iii

4 Table of Contents ACKNOWLEDGMENTS... III TABLE OF CONTENTS... IV LIST OF TABLES... VI LIST OF FIGURES... VII LIST OF ABBREVIATIONS... VIII CHAPTER 1 INTRODUCTION PANCREATIC CANCER Pancreatic Cancer Therapy Development Pancreatic Carcinogenesis OXIDATIVE STRESS Hypoxia The Impact of High ROS Levels Adaptations to High ROS Levels The Impact of Low ROS Levels THE ROLE OF ANTIOXIDANTS The Glutathione Antioxidant System The Thioredoxin Antioxidant System A NEW THERAPEUTIC STRATEGY FOR PANCREATIC CANCER CHAPTER INTRODUCTION iv

5 2.2 MATERIALS & METHODS RESULTS DISCUSSION CHAPTER OVERALL DISCUSSION BSO potentiates the acute cytotoxicity of auranofin Apoptosis contributes to cell death The drugs engage their expected targets Oxidative stress follows combination therapy IMC is an effective tool to examine auranofin in vivo FUTURE DIRECTIONS CONCLUSIONS REFERENCES v

6 List of Tables Table 2.1. IC50 values of auranofin monotherapy and in combination with BSO in primary pancreatic cancer cell lines Table 2.2. Panel of markers measured by IMC using subcutaneous PDX tumours Table 2.3. Estimation of the mean ions/pixel in the stroma and epithelium of subcutaneous tumours from auranofin treated PDX mice vi

7 List of Figures Figure 2.1. Dose-response of antioxidant pathway inhibitors, TrxR isozyme levels, and colony formation potential Figure 2.2. Representative two parameter plots measuring cell death and apoptosis in OCIP Figure 2.3. Representative two parameter plots assessing the GSH content of apoptotic cells in OCIP Figure 2.4. Effect of treatment on thiol and ROS levels in 9A and OCIP23 cell lines Figure 2.5. Auranofin-TrxR interaction in vitro Figure 2.6. Distribution of auranofin in subcutaneous PDX tumours vii

8 List of Abbreviations 5-FU ARE 5-fluorouracil antioxidant response element ASK1 apoptosis signaling kinase 1 Bax bcl-2-like protein 4 BCNU BSA BSO c-flip CETSA DCF DMEM DMSO ECM EGFR ERK FBS FFPE G6PD GCL GCLC GCLM GGT carmustine bovine serum albumin buthionine sulfoximine cellular FLICE-inhibitory protein cellular thermal shift assay 2,7 -dichlorofluorescein Dulbecco s Modified Eagle Medium dimethyl sulfoxide extracellular matrix epidermal growth factor receptor extracellular signal-regulated kinase fetal bovine serum formalin-fixed, paraffin-embedded glucose 6-phosphate dehydrogenase glutamate-cysteine ligase glutamate-cysteine ligase catalytic subunit glutamate-cysteine ligase modifier subunit -glutamyltranspeptidase viii

9 GLUD1 GOT1 GPx GR Grx GS GSH GSSG GST HNE IKK- IMC glutamate dehydrogenase aspartate transaminase glutathione peroxidase glutathione reductase glutaredoxin glutathione synthetase glutathione oxidized glutathione glutathione S-transferase 4-hydroxynonenal inhibitor of nuclear factor kappa-b kinase subunit beta imaging mass cytometry Keap1 kelch-like ECH-associated protein 1 mbbr MDH1 monobromobimane malate dehydrogenase ME1 malic enzyme 1 MMP Msr NADPH mitochondrial membrane potential methionine sulfoxide reductase nicotinamide adenine dinucleotide phosphate Nrf2 nuclear factor, erythroid derived 2, like 2 P-gp multidrug resistance protein 1 PanIN PBS pancreatic intraepithelial neoplasia phosphate-buffered saline PD-L1 programmed death ligand 1 ix

10 PDX PEGPH20 PHGDH PI PPP Prx PTEN RNR ROS RPMI RT SCID System XC TRAIL Trx TrxR VEGF-A patient-derived xenograft pegvorhyaluronidase alfa phosphoglycerate dehydrogenase propidium iodide pentose phosphate pathway peroxiredoxin phosphatase and tensin homolog ribonucleotide reductase reactive oxygen species Roswell Park Memorial Institute room temperature severe combined immunodeficiency cystine/glutamate antiporter system TNF-related apoptosis-inducing ligand thioredoxin thioredoxin reductase vascular endothelial growth factor A x

11 1.1 Pancreatic Cancer Chapter 1 Introduction Pancreatic cancer has the worst prognosis amongst all cancers; 5 year survival rates for the disease range from less than 5 to 8% 1 3. Although it is only the twelfth most commonly diagnosed, it is the fourth most common cause of cancer death in Canada 2. Unfortunately, mortality from pancreatic cancer has not improved during the era of modern medicine 4,5. Estimates predict that pancreatic cancer will become the second most common cause of cancer death behind non-smallcell lung cancer by The dismal outlook for patients with pancreatic cancer can be partially attributed to the lack of progress made in the development of diagnostics and therapeutics, which have remained stagnant over the last 50 years 7. During this time vast improvements have been made for the diagnosis and treatment of other tumour types. For example, the development of the Pap smear can reduce cervical cancer death by up to 80% 8, and the development of trastuzumab has increased overall survival by 33% in HER2+ breast cancers 9,10. The identification of new biomarkers and treatment strategies is required to overcome pancreatic cancer. Early stage pancreatic cancer is generally asymptomatic, leading to difficulties in detection. The disease may be identified secondary to non-specific symptoms, or as an incidental finding. In addition to this, screening tests have not been developed. Currently the only treatment that offers the potential for long-term survival is surgery in early-stage disease, however the 5 year survival rate in patients with resectable tumours remains only 15-20% 11. For patients with advanced metastatic disease there are few systemic options that offer little benefit in survival. 1

12 Pancreatic Cancer Therapy Development The first systemic therapy used in the treatment of pancreatic cancer was 5-fluorouracil (5-FU). 5- FU is a non-specific agent that inhibits the synthesis of the nucleoside thymidine. 5-FU provided a 4 month survival benefit when used in combination with radiotherapy, compared to radiotherapy alone 12. Since the indroduction of 5-FU in pancreatic cancer clinics some small advancements have been made. Gemcitabine became the new standard in the late 1990s when a clinical trial showed it offered a significant improvement in median overall survival when compared to 5-FU (5.6 months vs. 4.4 months) 13. However, gemcitabine is another non-specific agent. It is a nucleoside analog that replaces cytidine during DNA replication. Since the introduction of gemcitabine in the treatment of pancreatic cancer, two therapies have been developed for clinical use. The combination therapy FOLFIRINOX (folinic acid, 5-FU, irinotecan, and oxaliplatin) improved median overall survival compared to gemcitabine in a clinical trial (11.1 months vs. 6.8 months) 14, and the combination of nab-paclitaxel with gemcitabine improved median overall survival compared to gemcitabine alone (8.5 months vs. 6.7 months) 15. Although there appears to be a survival difference between FOLFIRINOX and gemcitabine/nab-paclitaxel, the FOLFIRINOX regimen is more toxic and the patient population was skewed slightly in favour of the FOLFIRINOX cohort, leading many investigators to consider the two regimens to be similar Although these therapies are an improvement on past therapies, median survival falls short of one year with either regimen. Great interest has centred on the development of targeted therapies to provide effective non-toxic treatment for pancreatic cancer. There have been many proposed targets and clinical trials 19, however no such therapy has entered clinical practice in Canada to date. Trials have examined agents such as bevacizumab, inhibiting angiogenesis by targeting vascular endothelial growth

13 3 factor A (VEGF-A), or cetuximab, inhibiting epidermal growth factor receptor (EGFR), in combination with gemcitabine to no benefit 20,21, among many other attempts. From the trials, only erlotinib in combination with gemcitabine has shown a significant effect on overall survival (6.2 months vs. 5.9 months), although this is not clinically relevant 22. In summary, current attempts at targeted therapy have proven ineffective in the management of pancreatic cancer. This may change in the future, however current prospects seem bleak and new biomarkers may be required to develop these therapies further 23. Beyond targeted therapy there have been attempts to develop immunotherapies for pancreatic cancer. Immunotherapy is a treatment approach that relies on augmenting the host s immune response against cancer 24. This method derives from the observation that the immune system can eliminate malignant cells during the early stages of transformation 25. While there have been many unequivocal successes in other tumour types, particularly in advanced melanoma 26 28, pancreatic cancer has proven resistant to these approaches. Clinical trials have observed a lack of response to immune checkpoint blockade with ipilimumab 29 and programmed death ligand 1 (PD-L1) directed monoclonal antibodies 30. These failures have resulted from the poorly immunogenic and immunosuppressive nature of the pancreatic cancer microenvironment 7,31. A small number of patients enrolled in these trials have received some therapeutic benefit, leading to continued interest in developing new immunotherapy strategies. One such strategy includes attempting to achieve a synergistic response through the use of multiple agents. This may include combining immune checkpoint blockade, inhibition of immunosuppressive signalling, stimulation of T cells, and/or conventional cytotoxic therapies 32. In relation to immunotherapy, other strategies attempt to target additional features of the tumour microenvironment. Attention has focused on the role of the stroma in determining therapy

14 4 response. These tumours are characterized by a dense desmoplastic reaction comprised of fibrotic tissue, extracellular matrix proteins, pancreatic stellate cells, immune cells, and blood vessels 7. The extensiveness of desmoplasia correlates with worse disease outcome 33. Furthermore, desmoplasia limits tumour vascularization, creating a barrier to drug uptake by tumours 34. A clinical trial using the modified hyaluronidase molecule pegvorhyaluronidase alfa (PEGPH20) to degrade stromal hyaluronan in combination with nab-paclitaxel and gemcitabine has shown improvement in progression free survival by the perceived mechanism of increasing tumour vasculature to aid drug delivery 35. However, there is evidence to suggest that the stroma may have a role in protection against tumour progression and an increased tumour vasculature may contribute to accelerated growth if not targeted properly 7. Therefore, there is a difficult balance to achieve via this strategy. Clinical trials have also given attention to hypoxia, which is a feature of solid tumours that can result from the lack of vasculature in the stroma. Hypoxia is a tumour microenvironment state where oxygen utilization exceeds its supply. Traditionally, pancreatic tumours have been considered to be characterized by extensive hypoxia, which may be caused by the lack of blood supply resulting from the desmoplastic reaction 36. In patient-derived xenograft (PDX) mouse models, hypoxia correlates with aggressive growth and metastasis 37, making it a logical choice to target. The prodrug evofosfamide is activated under hypoxic conditions to release a DNA alkylating agent that inhibits cancer cell replication 38. However, a phase III clinical trial observed no significant improvement in median overall survival using evofosfamide/gemcitabine compared to gemcitabine alone (8.7 months vs. 7.6 months) 38. It is possible that hypoxia in patient tumours is more variable than previously believed, which would limit the effectiveness of therapy across the entire patient population 39. This can be addressed by non-invasive imaging using positron-emitting hypoxia tracers in order to stratify patients towards hypoxia-targeting clinical trials 40.

15 5 In summary, the most effective systemic therapies available for metastatic pancreatic cancer provide only a small benefit to patients. Although there are promising avenues to explore further, attempts to develop new strategies against pancreatic cancer have fallen short of expectations. Further studies are required to identify and examine novel, exploitable targets in pancreatic cancer Pancreatic Carcinogenesis Four driver genes are frequently mutated in pancreatic cancer. These are the oncogene KRAS and the tumour suppressors CDKN2A, SMAD4, and TP These mutations are shared by pancreatic intraepithelial neoplasia (PanIN), which is a precursor lesion to pancreatic cancer. Sequencing of these genes in PanINs suggests a stepwise progression of pancreatic cancer. Activation of KRAS is present in the earliest PanIN-1 lesions, making it the initiating event in pancreatic cancer tumorigenesis. Inactivation of CDKN2A is found in PanIN-2, followed by inactivation of SMAD4 and TP53 in PanIN-3 lesions 45. However, more recent whole-genome sequencing of tumour cell-enriched primaries showed cases that displayed either simultaneous inactivation of multiple drivers or tumours that did not follow the predicted mutation sequence. These data are consistent with punctuated rather than gradual evolution 46. Regardless, KRAS activation occurs in more than 90% of human pancreatic cancer and its effects likely have the greatest impact on pancreatic cancer 41. Interestingly, KRAS has been implicated in mediating pancreatic carcinogenesis in part through its impact on oxidative stress. Its most important effect on oxidative stress is the promotion of low intracellular reactive oxygen species (ROS) levels by inducing nuclear factor, erythroid derived 2, like 2 (Nrf2) activation 47. This aspect will be discussed further in the next section.

16 6 1.2 Oxidative Stress Pancreatic cancer, among other cancers, is characterized by excessive oxidative stress 48. Oxidative stress is caused by the overproduction of ROS, leading to damage of lipids, proteins, and DNA 49. It occurs when there is an imbalance between intracellular ROS and antioxidants in favour of the oxidants. ROS are produced at an extreme rate due to the high metabolism inherent in transformed cells, and may promote aggressive tumour features 50,51. Oxidative stress and ROS represent an interesting target in pancreatic cancer because ROS exhibit dual roles; lower concentrations of ROS can stimulate cancer cell proliferation, whereas higher concentrations of ROS are damaging to cells 49,52. Cancer cells adapt during oxidative stress conditions to maintain low ROS levels to avoid cell death and promote proliferation by enhancing antioxidant defences 53. ROS exist as multiple species including superoxide (O2 ), hydroxyl radical (HO ), and hydrogen peroxide (H2O2) 54. They are produced in various subcellular compartments: the cytoplasm, the mitochondria, peroxisomes, and the endoplasmic reticulum. The major source of cytoplasmic ROS in pancreatic cancer is NADPH oxidase, which produces superoxide anion by facilitating the reduction of molecular oxygen by NADPH 55. Mitochondrial ROS are produced in dysfunctional mitochondria that result from inefficient hypermetabolism in rapidly proliferating cancer cells. This can cause leakage of electrons at complex I and complex III of the mitochondrial electron transport chain, leading to the production of superoxide and its dismutation to hydrogen peroxide by superoxide dismutases 56,57. Peroxisomal ROS are produced during the -oxidation of fatty acids 58. Lastly, endoplasmic reticulum can contribute to ROS production through its role in providing an oxidizing environment to promote disulfide bonding and protein folding 59. Although ROS themselves are short-lived molecules, they can generate longer lasting lipid peroxides such as 4-hydroxynonenal (HNE) 60. HNE is capable of reacting with cellular macromolecules such as

17 7 proteins, lipids, and DNA. In addition, HNE can stimulate signaling cascades to affect cell proliferation and differentiation 61. As such, it has been implicated in different cancers Hypoxia Hypoxic tumours experience both chronic and acute hypoxia. Chronic hypoxia exists in regions located away from blood vessels and is limited by the diffusion of oxygen through the dense tumour stroma. Acute or cycling hypoxia exists in regions proximal to blood vessels and is perfusion-limited in that there are temporal instabilities in blood flow through a tumour. Cycling hypoxia generates greater oxidative stress because the acute influx of oxygen can contribute to an increase in ROS 62. In pancreatic xenografts hypoxia is associated with aggressive growth and metastasis 37. However, human pancreatic tumours may not possess extensive hypoxia as previously suggested 40. Although hypoxia stems from a lack of oxygen, hypoxic tumours have been stated to overproduce ROS 54. However, direct evidence suggesting this phenomenon is lacking. Hypoxia can contribute to resistance to chemo and radiotherapy, which rely on ROS to mediate their effects 63. This evidence contradicts the idea hypoxia of greater oxidative stress during hypoxia The Impact of High ROS Levels At high levels ROS disrupt cellular processes by non-specifically damaging proteins, lipids, and DNA. Furthermore, ROS are implicated in both the extrinsic (death receptor-dependent) and intrinsic (mitochondrial) pathways of apoptosis. The extrinsic pathway is mediated by deathinducing ligands, such as TNF-related apoptosis-inducing ligand (TRAIL) 64. Following ligandreceptor interaction the caspase cascade is activated, leading to apoptosis 65. ROS negatively regulate cellular FLICE-inhibitory protein (c-flip) by a post-transcriptional method to target it for degradation and activate the extrinsic apoptosis pathway. The intrinsic pathway is mediated by

18 8 a permeability transition in the mitochondria. Mitochondrial stress, including oxidative stress, stimulates pore formation in the inner mitochondrial membrane which causes cytochrome-c mobilization and mitochondrial swelling 66. This can lead to rupture of the outer membrane, and with the help of pro or anti-apoptotic Bcl-2 family member proteins can stimulate or prevent apoptosis. Bcl-2-like protein 4 (Bax) in particular mediates the formation of the mitochondrial apoptosis-induced channel (MAC) to release mitochondrial pro-apoptotic factors, such as cytochrome-c, which leads to activation of the executioner caspase proteins As mitochondria are a significant source of ROS production, the intrinsic pathway is the largest contributor to ROSmediated apoptosis. In this pathway, electron transport chain-produced superoxide triggers eventual cytochrome-c release, leading to the induction of apoptosis 70. Apoptosis following oxidative stress displays several distinct steps that can be measured by flow cytometry methods. Following glutathione (GSH) depletion in cells, ROS production causes the mitochondrial permeability transition and the subsequent loss of mitochondrial membrane potential (MMP), due to cellular damage. Finally, cells lose viability after perforation of the cell membrane 71. In addition to apoptosis, ROS contribute to ferroptosis, a form of cell death that depends on intracellular iron 72. NADPH oxidase-induced accumulation of ROS was identified to contribute to ferroptosis. Furthermore, the compounds erastin and sulfasalazine, which inhibit the cystine/glutamate antiporter system (system XC ), lead to toxic accumulation of ROS and ferroptosis 72,73. The lethal effect can be reversed by addition of the lipophilic antioxidants ferrostatin-1 and vitamin E, suggesting an important role for lipid radicals in mediating ferroptosis 72,74. KRAS driven cancer cells have been shown to be susceptible to ROS-mediated ferroptosis following treatment with lanperisone, an FDA-approved muscle relaxant 75, suggesting that ROS can be utilized to induce ferroptosis in a KRAS driven tumour such as pancreatic cancer.

19 9 High ROS also have indirect effects on cell death by sensitizing cancer cells to chemo and radiotherapy. The multidrug resistance protein 1 (P-gp) facilitates the removal of several anticancer agents from cells. High ROS concentrations have been associated with decreased P-gp expression, which allows for cytotoxic drugs to accumulate within a cell 76. Furthermore, ROS have been identified to mediate the effects of some cytotoxic chemotherapies. For example, gemcitabine acts through a ROS-dependent mechanism that may promote metastasis due to the action of antioxidants 77. This feature may contribute to gemcitabine s limited effectiveness in pancreatic cancer. Moreover, both 5-FU and oxaliplatin have been observed to augment oxidative stress in patients with colon cancer 78. These drugs are two of the components of the combination FOLFIRINOX, which is one of the most effective systemic therapies against metastatic pancreatic cancer. Paclitaxel, which is the main component of the nab-paclitaxel/gemcitabine therapy for pancreatic cancer, and doxorubicin also enhance oxidative stress 79,80. While augmented oxidative stress is not the primary mechanism of action of these therapies, it is interesting to note that the most effective therapies in clinical use for pancreatic cancer elicit an effect on the redox environment of the tumour. In addition to chemosensitization, high ROS levels are involved in mediating the cytotoxic effects of radiotherapy. Radiotherapy utilizes ionizing radiation to generate ROS, particularly hydroxyl radical. Hydroxyl radicals enact the therapeutic and side effects of radiotherapy. It propagates by reacting with other molecules including DNA, leading to abundant DNA lesions and subsequent cell death 81. It has been observed that radical scavengers are effective at attenuating the cytotoxic effects of radiation, indicating that ROS are required The evidence presenting the role of ROS in chemo and radiosensitization suggests that exploiting an increase in ROS can be effective in the development of new therapies.

20 10 Cancer cells require multiple adaptations to protect from inherently high ROS levels allowing them to survive and proliferate. Signaling through the oxidative stress response mediates this adaptation. The master regulator of this response is the transcription factor Nrf2. Nrf2 s activity is modulated by its repressor Kelch-like ECH-associated protein 1 (Keap1). In a normal redox state Nrf2 is constitutively degraded. Keap1 promotes its ubiquitination by Cullin E3 ubiquitin ligase to label Nrf2 for proteasome degradation 85. During oxidative stress conditions, redox-sensitive cysteine residues on Keap1 are oxidized by ROS. In this state Keap1 is unable to bind Nrf2 to recruit ubiquitin ligases to target it for degradation. Nrf2 subsequently translocates to the nucleus where it heterodimerizes with small Maf proteins to stimulate transcription of genes associated with antioxidant response element (ARE) promoters 86. The lipid peroxide HNE is a potent inducer of Nrf2 by covalent binding to Keap1 redox-sensitive cysteines and by activating upstream kinases such as protein kinase C 87,88 It has been observed that nuclear Nrf2 expression associates with poor survival in pancreatic cancer 89, whereas cytoplasmic Keap1 expression associates with reduced metastasis and membrane-associated Keap1 displays a protective effect on survival Adaptations to High ROS Levels Nrf2 and ARE positively regulate the transcription of several cytoprotective genes including drugmetabolizing enzymes, drug transporters, and antioxidants 91. Nrf2 regulates the expression of many phase I and II drug-metabolizing enzymes 92 94, as well as members of the multi-drug resistance-associated gene family of multidrug transporters 95. It can indirectly modulate ROS levels by regulating free Fe (II) homeostasis. This prevents the Fenton reaction, where Fe(II) catalyses the oxidation of hydrogen peroxide to the more reactive hydroxyl radical 96. This function occurs by promoting the release of Fe (II) from heme, and its subsequent sequestration. However, Nrf2 s most important role is functioning as the master regulator of antioxidants. The production

21 11 and regeneration of the most abundant antioxidant cofactor, GSH, is controlled exclusively by Nrf2 97. In addition, Nrf2 supports the expression of enzymes involved in GSH utilization 98, as well as enzymes involved in the production, regeneration, and utilization of thioredoxin (Trx) 99. These actions of Nrf2 are responsible for direct modulation of oxidative stress. These antioxidants will be discussed in further detail in section 1.3. In parallel to antioxidant production, cancer cells require the redirection of cellular metabolism to support antioxidant function. Cancer cell metabolism differs from that of normal cells and requires abundant ATP production to support rapid proliferation. Cancer cells experience a metabolic switch to rely on aerobic glycolysis, rather than oxidative phosphorylation, to supplement their needs. This phenomenon is termed the Warburg effect 100,101. Constitutively active KRAS in pancreatic cancer plays a key role in activating the metabolic switch towards supporting anabolic pathways 102. In addition to this switch, Nrf2 reprograms metabolism in oxidative stress conditions to support the antioxidants it promotes. It reprograms metabolism by increasing transcription of NADPH-generating enzymes to modulate glucose, glutamine, and serine metabolism 103,104. NADPH is an important cofactor that is required by antioxidant enzymes to supply reducing potential. Nrf2 promotes a proliferative phenotype through the PI3K-Akt axis, which redirects glucose into anabolic pathways to increase metabolism 103. Three major pathways that produce NADPH are regulated by Nrf2: the pentose phosphate pathway (PPP), the serine synthesis and folate pathway, and the malic enzyme pathway of glutamine metabolism. The PPP is a metabolic pathway that runs in parallel to glycolysis. It consists of oxidative and nonoxidative phases. The oxidative arm is responsible for NADPH production, while the nonoxidative arm is responsible for the generation of ribonucleotides 105. Both phases are critical to cancer cell survival. Glucose 6-phosphate dehydrogenase (G6PD) is the first enzyme in the

22 12 oxidative arm of the PPP and is positively regulated by Nrf2 93. G6PD commits glucose 6- phosphate to the PPP, rather than glycolysis 105. The PPP has been observed to promote cell survival and proliferation during neoplastic transformation 106. Although the oxidative phase has generally been implicated to supply cancers with NADPH, constitutively active KRAS has been observed to uncouple ribose biogenesis from NADPH production in pancreatic cancer 107. Glycolytic intermediates are utilized specifically in the non-oxidative arm of the PPP, but not the oxidative arm. Inhibition of the PPP by knockdown of G6PD, as well as glucose deprivation, in pancreatic cancer cells led to minimal effect on ROS, indicating that the oxidative arm is not required to maintain ROS levels in pancreatic cancer 108. ROS could be affected, however, by modulating the other two NADPH production pathways. The serine synthesis pathway is another glycolytic shunt responsible for NADPH production. In this pathway the glycolytic intermediate 3-phosphoglycerate is converted to serine via three reactions. The first committed step is catalyzed by phosphoglycerate dehydrogenase (PHGDH). After its production serine can be used in the folate cycle to generate glycine and NADPH 109. The transcription of the enzymes involved in this process is regulated by Nrf Overexpression of PHGDH has been reported in non-small cell lung, colorectal, breast, and pancreatic cancers 104, In pancreatic cancer, PHGDH has been described as an independent prognostic indicator for patients; increased expression is associated with poor prognosis. Furthermore, PHGDH promotes proliferation, migration, and invasion of pancreatic cancer cells in vitro 112. Reduced levels of serine hydroxymethyltransferase, the enzyme responsible for preparing serine for the folate cycle, were responsible for decreasing NADPH while increasing ROS 113. Therefore, Nrf2 promotes the serine synthesis and folate pathway to produce NADPH, which allows the cancer cells to adapt to oxidative stress and may contribute to worse outcomes in pancreatic cancer.

23 13 Finally, glutamine metabolism through the malic enzyme (ME1) pathway is another source of NADPH production that may be the most important in pancreatic cancer 108. ME1 is positively regulated by Nrf2 99. Cancer cells exhibit glutamine addiction because of its essential canonical roles in supplying intermediates for the Kreb s cycle through anaplerosis, and its roles in nucleotide, amino acid, and lipid biosynthesis Most cells utilize the enzyme glutamate dehydrogenase (GLUD1) to convert glutamine-derived glutamate into -ketoglutarate to fuel the Krebs cycle, however constitutively active KRAS in pancreatic cancer downregulates GLUD1 and upregulates aspartate transaminase (GOT1) to shunt glutamine into the non-canonical ME1 pathway for NADPH production 108. In this pathway glutamine-derived aspartate is converted to oxaloacetate via GOT1 in the cytoplasm, and then oxaloacetate is metabolized by malate dehydrogenase (MDH1) into malate. The final step involves the oxidation of malate to pyruvate by ME1. This reaction is responsible for NADPH regeneration and assists in protecting from damage during oxidative stress 117. Confirming this, glutamine deprivation and ME1 knockdown led to elevated ROS and decreased clonogenicity in pancreatic cancer cells. The cells could be rescued with supplemented oxaloacetate, malate, or GSH, indicating that glutamine is required for redox homeostasis in pancreatic cancer 108. Further demonstrating the importance of glutamine metabolism in pancreatic cancer, ME1 pathway inhibition diminished tumour xenograft growth The Impact of Low ROS Levels The adaptations controlled by Nrf2 maintain ROS at low to moderate levels (ie. above background levels but below cytotoxic levels) to prevent cell death. These low ROS levels are implicated in stem cell phenotype and in tumour cell proliferation and act as important signaling molecules capable of regulating various functions including proliferation, apoptosis, and tumour cell invasion 48. Low ROS levels are associated with maintaining cancer stem cell-like properties in

24 14 cells that confers therapy resistance 118. For example, human breast cancer stem cells possess robust ROS scavenging systems that maintain low ROS levels 84. This feature prevents ROS-induced DNA damage following ionizing radiation, protecting the cells from death. Furthermore, pharmacologic depletion of antioxidants results in loss of stem cell properties 84. Low ROS are involved in signal transduction pathways to regulate various cellular responses. This function is made possible because of ROS-mediated inactivation of protein-tyrosine phosphatases, which allows for increased signaling through receptor tyrosine kinases 119. As such, ROS activate signaling of the MAPK pathway by activating p38 MAPK to promote tumour cell differentiation and proliferation 120. In addition, ROS can inactivate PI3K/Akt phosphatases like PTEN. This stimulates signaling of the PI3K pathway to promote proliferation and apoptosis evasion 55,121. Proliferation promotion by ROS has been observed in various cancer cell types 122,123. These functions implicate ROS in carcinogenesis. Moreover, ROS may contribute to the promotion of invasion and metastasis. Tumour cell invasion and metastasis requires cell mobility through the extracellular matrix (ECM). Cells degrade glycosaminoglycan in the ECM to allow for motility. ROS-dependent degradation of the ECM occurs due to the stimulation of matrix metalloproteinase activity, expression, and secretion 124. ROS-activated MAPK pathways contribute to the expression of matrix metalloproteinases. Specifically, extracellular signal-regulated kinase (ERK) activation by ROS promotes invasion and metastasis 125. The metastatic phenotype is enabled and sustained by concomitant ROS production and the ability to protect from the cytotoxicity of oxidative stress with antioxidants to maintain low ROS levels 126. These functions of low ROS levels allow for tumour growth and spread that may contribute to worse prognosis in pancreatic cancer. Taken together this information indicates that low ROS levels contribute to many of the tumorigenic properties of cancer cells. Cancer cells utilize antioxidants to protect themselves from

25 15 ROS and maintain these low ROS levels. These antioxidants are regulated by the master regulator Nrf2 during the oxidative stress response. ROS scavenging prevents ROS-induced toxicity and assists in ROS-mediated proliferation. Strategies designed to exacerbate ROS through targeted inhibition of antioxidants may provide a relatively non-toxic therapy, as cancer cells are reliant on antioxidants whereas normal cells can survive change. This strategy has been employed using various drugs against a number of targets to great success 75,127, The Role of Antioxidants The most important factor in the cell that moderates ROS levels are the antioxidants regulated by Nrf The role of antioxidants and Nrf2 in cancer is controversial as studies suggest them to be either tumour suppressive or oncogenic. The answer to this controversy is likely to be that the role of antioxidants is context specific 91,130. Nrf2-knockout mice are more susceptible to carcinogenesis and show diminished efficacy of chemopreventive enzymes 131. Furthermore, the natural compound sulforaphane activates Nrf2 and was found in a phase II clinical trial to protect against hepatocellular carcinoma by preventing DNA adduct formation 132. Although these studies suggest that the antioxidant response mediated by Nrf2 is required for chemoprevention, most clinical trials suggest that dietary antioxidants do not provide this effect 133,134, and may even increase cancer risk 135,136. Nrf2 may promote tumorigenesis by protecting cells from oxidative stress and maintaining low ROS. Oncogenes such as KRAS in pancreatic cancer increase Nrf2 to provide cytoprotection and decrease ROS levels 47. Loss-of-function mutations in KEAP1, resulting in constitutive Nrf2 activation, have been identified in various solid tumours 137. Lastly, high Nrf2 expression has been correlated with resistance to platinum-based chemotherapy in ovarian cancer 138.

26 16 These contrasting sets of data suggest that the role of Nrf2 may depend on the stage of tumorigenesis. Before the cancer development, Nrf2 and antioxidants may help to suppress cancer by the prevention of mutations and oxidative stress. Once the cancer develops, Nrf2 and the antioxidants it regulates may promote cancer progression 91,130. Therefore, once a tumour has been developed, antioxidants may prove to be an important target for cancer therapy. As previously discussed, the GSH and Trx systems are both coordinated by Nrf2. These are likely two of the main factors contributing to tumour cytoprotection. Not only is the production of GSH and Trx controlled by Nrf2, but also the production of the enzymes that recycle and utilize them to evoke their effects The Glutathione Antioxidant System GSH is the most abundant non-protein thiol in cells and is critical to providing protection from oxidative stress 139. GSH exists in two forms: reduced GSH, or oxidized glutathione (GSSG) which is a dimer. A number of enzymes responsible for GSH homeostasis are regulated by Nrf2 during periods of oxidative stress. GSH synthesis is performed in a two-step process from the amino acids cysteine, glutamate, and glycine. The first reaction in GSH biosynthesis is the rate-limiting step catalyzed by glutamate-cysteine ligase (GCL), which has both a catalytic (GCLC) and modifier (GCLM) subunit 140. GCLC performs the catalytic activity of the enzyme 141, while GCLM is an important regulatory subunit that lowers the Km for glutamate 142. In this reaction glutamate and cysteine are ligated to form -glutamylcysteine. This dimer contains a non-conventional peptide bond between the -carboxyl group of glutamate and the amine of cysteine, rather than the - carboxyl group. Hydrolysis of this bond can only be performed by -glutamyltranspeptidase (GGT) which is found on the external surface of certain cell types and supplies cells with GSH precursors 143. As such GSH resists intracellular degradation. The second reaction to add glycine

27 17 to -glutamylcysteine is catalyzed by GSH synthetase (GS). Overexpression of GS fails to increase GSH levels, whereas GCL overexpression leads to an increase in GSH 144, supporting the notion that GCL is more important in the regulation of GSH synthesis. Since -glutamylcysteine is found at very low intracellular concentrations, GCL is considered to be rate-limiting 140. Cystine uptake through the system XC transporter can also be a limiting step in GSH synthesis, as intracellular cysteine concentrations are low 145. Cysteine auto-oxidizes extracellularly to cystine, which can then be taken up by the system XC transporter before rapid intracellular reduction 145. The expression of both GCL subunits and the critical xct subunit of the system XC are regulated by Nrf2 146, Functions of the Glutathione System As a redox sensitive thiol the primary function of GSH is to scavenge ROS 148. Supporting the antioxidant role of GSH, depletion of GSH leads to increases in ROS and subsequent ROS-induced toxicity 149. GSH is particularly required in the mitochondria to protect from pathologically generated oxidative stress 150. This antioxidant function is performed in reduction reactions catalyzed by glutathione peroxidase (GPx). GPx catalyzes the reduction of hydrogen peroxide and lipid peroxides, which in turn oxidizes GSH to GSSG 151. Members of the GPx enzyme family, including GPx2 and GPx4, are regulated by Nrf2 94,152. GSH recycling is performed when cells reduce GSSG using the enzyme glutathione reductase (GR) and NADPH to maintain a reducing environment. This key enzyme for maintaining GSH homeostasis is regulated by Nrf Since the GSH to GSSG ratio is a significant contributor to the redox equilibrium, cells require substantially more GSH than GSSG to maintain a reducing environment. During periods of oxidative stress when GSSG accumulates, cells can export GSSG when the capacity of GR to recycle GSH is outmatched 154. Another mechanism to prevent toxic GSSG accumulation during

28 18 oxidative stress is the reaction of GSSG with a protein thiol catalyzed by protein disulfide isomerase to form a protein mixed disulfide 155. GSH can contribute to redox signaling in addition to its antioxidant function. This role is executed by modifying the oxidation state of critical modifying cysteine residues in proteins 156. Mechanistically this is performed by the reversible binding of GSH to protein thiols to alter the protein structure and function. This process is called glutathionylation and can activate or inactivate proteins 157. Glutathionylation can happen enzymatically with the assistance of GSH S- transferase (GST) or non-enzymatically 158,159. This mechanism exists to protect sensitive proteins from irreversible oxidation. Deglutathionylation is catalyzed by the oxidoreductase glutaredoxin (Grx) which is itself reduced by GSH 160. A significant number of proteins critical in cell signaling have been identified to be regulated by glutathionylation. These include GAPDH, inhibitor of nuclear factor kappa-b kinase subunit beta (IKK- ), and p This process may affect ROS generation, as glutathionylation of electron transport chain proteins can alter their function 165. GST has an additional role in phase II detoxification of xenobiotics, including chemotherapeutic drugs. In this process GSH is conjugated to xenobiotics during metabolism to prepare for excretion 166. GST-mediated detoxification can contribute to chemoresistance 167. GSH contributes to the regulation of growth in both normal and transformed cells. Increased GSH levels are essential for cell cycle progression and are associated with proliferation Hepatocytes increase GSH in response to injury, as has been observed during liver regeneration following partial hepatectomy 171. Blocking this GSH generation causes a reduction in the total amount of DNA synthesis associated with cell division 170. GSH levels in various cancer types including hepatocellular carcinoma, melanoma, and pancreatic cancer have been correlated with cell and tumour growth 170,172,173. Interestingly, the redox environment of the nucleus has been

29 19 observed to be critical during the cell cycle. During proliferation GSH localizes to the nucleus and can affect gene expression of multiple proteins through influence on histones 174. Cell cycle progression requires a reducing environment moderated by GSH to occur. In addition to cell growth regulation, GSH is also involved in cell death regulation. This regulation is carried out through effects on the mechanism of apoptosis. GSH levels impact the expression and activity of caspases that are involved in effecting apoptosis 175. As previously mentioned, GSH levels fall as ROS propagate during the apoptotic cascade 149. This can occur because of decreased GCL activity following GCLC cleavage by caspase 3 during apoptosis Targeting the Glutathione System in Cancer GSH is critical for malignant behaviour of cancer cells. GSH is required during tumour initiation, as its depletion impairs tumorigenesis 127. This is because GSH prevents oxidative stress from enacting its cytotoxic effects on transformed cells. Following tumour formation GSH becomes important to maintain high growth rates, as GSH levels are correlated to growth 177. These levels are elevated in cancer due to increased expression of GCL and the system XC 178,179. In addition, GGT is an early marker of neoplastic transformation and assists with the GSH cycle to supply cancer cells with GSH metabolic precursors 180. GGT overexpression has been observed in melanoma and hepatocellular carcinoma 181,182, among other cancers. Lastly, GSH is required for establishing distant metastases because metastatic cells experience higher oxidative stress. In response they elevate GSH levels to combat the microenvironment and maintain redox homeostasis 126. To highlight the importance of GSH in cancer, many studies have attempted pharmacologic GSH depletion and observed sensitization to various therapies. Cancer cells are sensitized to radiation upon GSH depletion 84. In addition, chemoresistance has been associated with high GSH levels in multiple myeloma and ovarian cancer 165,166. This chemoresistance can be eliminated following depletion of GSH 184

30 20 As GSH is critical to cancer progression and survival there is reason to believe that it can be exploited in cancer therapy. Various key nodes in GSH metabolism can be targeted as an anticancer strategy. First, substrate supply for GSH biosynthesis can be targeted. Glutamine analogs targeting GGT have been evaluated in humans but have been found too toxic for clinical use 185. Newer uncompetitive inhibitors are in development but have yet to prove effective and will require time before clinical application 186. The system XC transporter may be a critical target of an anti-gsh cancer therapy as it is required to regulate intracellular cysteine and has been observed to be necessary for cancer cell survival 187. Multiple inhibitors against this system have been described. Sulfasalazine is a drug that has been used in clinics for rheumatoid arthritis, ulcerative colitis, and Crohn s disease, and inhibits the system XC 188. However, the mechanism of action is poorly understood, it is metabolically unstable, and lacks potency 73. Erastin is a newer molecule that is more potent than sulfasalazine and more selective for system XC 74. However, it has yet to be tested in humans. A second potential target critical to GSH metabolism is GSH recycling. The nitrogen mustard compound carmustine (BCNU) is primarily an alkylating agent used in chemotherapy for brain cancer and lymphomas 189,190, however its off-target effect is GR inhibition 191. Unfortunately, BCNU is in short supply, costly, and displays prolonged myelosuppression and potentially lethal pulmonary toxicity 190,192,193. A newer molecule 2-AAPA has been described as a GR inhibitor that increases GSSG and stimulates protein glutathionylation 194. However, 2-AAPA has yet to be tested in humans and possesses off-target effects including Grx and thioredoxin reductase (TrxR) inhibition 195,196. These off-target effects and potentially others that have yet to be uncovered make 2-AAPA too unpredictable for use in cancer therapy.

31 21 Lastly, a third strategy involves direct targeting of GSH synthesis. GCL, the rate-limiting enzyme, is more important for GSH levels than GS, the enzyme catalyzing the second step of GSH synthesis 144. Therefore, GCL would be the ideal target. Fortunately, an irreversible inhibitor of GCL, buthionine sulfoximine (BSO), has been studied extensively in vitro, in vivo, and in human clinical trials 127,197,198. BSO has proven to be minimally toxic in humans when used on its own 199. Although it is rapidly metabolized following intravenous bolus, concentrations of ~500 M following continuous infusion are achievable in humans while maintaining safe toxicity profiles 200. BSO has garnered great interest for use in drug combinations against cancers that have been observed to require GSH for chemoresistance 197. Mechanistically, it contributes to oxidative stress in the cells by acute GSH depletion and long-term ROS exacerbation 149. This can lead to both apoptosis and ferroptosis 149,201. BSO s safety profile, specificity, low cost, previous clinical trials, and mechanistic understanding make it an excellent drug to develop further for cancer therapy that modulates oxidative stress. BSO on its own is not very cytotoxic with short-term treatments and many cell lines are resistant to acute GSH depletion 197. It is likely that therapy involving BSO would require a second agent to enhance toxicity. Adaptation to GSH depletion involves Nrf2 activation to upregulate other antioxidants, including Trx 202. Since antioxidants have similar, sometimes overlapping functions, compensation from alternative antioxidant systems is a plausible mechanism for resistance to GSH depletion. Compensation between antioxidants has been observed in other contexts including between GSH and Trx, and between Trx and heme oxygenase 127,203,204. Therefore, therapy tailored to inhibit more than one antioxidant system may be able to overcome compensation and produce strong anti-cancer effects.

32 The Thioredoxin Antioxidant System The Trx system is a reductase system that is required for DNA synthesis and protection from oxidative stress 203. The three major components of the system are Trx, TrxR, and NADPH. As with other antioxidant genes, Trx and TrxR are transcriptionally regulated during oxidative stress by Nrf2 99,205. Trx is a small (~12 kda) reductase that is the effector of the Trx system. It catalyzes the reduction of protein disulfides. The active site dithiol in reduced Trx interacts with various protein substrates via a thiol-disulfide exchange reaction that forms a disulfide in Trx (oxidized Trx). This oxidized form is reduced by electron transfer from NADPH via the reaction catalyzed by TrxR. There is a mitochondrial Trx2 with two active site cysteines, and cytosolic and nuclear Trx1 with 3 additional cysteines that contribute to enzymatic regulation 206. TrxR is a homodimeric oxidoreductase that recycles Trx, and belongs to the same protein family as GR 207. Mammalian cells contain three TrxRs: cytosolic TrxR1, mitochondrial TrxR2, and testis-specific TrxR Of these, TrxR1 has been identified as a fitness gene 209. Although structurally and functionally similar, the TrxR enzymes have different amino acid sequences 210. These enzymes contain a unique C-terminal selenocysteine residue that contributes to its reaction mechanism. In the mechanism of TrxR reduction electrons from NADPH are first transferred to enzyme-bound FAD, then FAD transfers reducing equivalents to the N-terminal disulfide in the redox centre of the same subunit. This reduces the disulfide to a dithiol, which then transfers the electrons to the C-terminal selenenylsulfide of the other subunit. Addition of another NADPH molecule reduces these residues to a selenol-thiol pair 211. The reduced enzyme can transfer these electrons to its substrates, including proteins such as oxidized Trx and Grx2, and small molecules such as hydrogen peroxide and cytochrome c 212. When transferring electrons to Trx the selenol of TrxR is deprotonated to a selenolate anion. This anion attacks the disulfide of oxidized Trx, resulting in a

33 23 TrxR-Trx mixed selenenylsulfide. The free thiol on the C-terminal TrxR reduces this bond, causing the reformation of the selenenylsulfide in TrxR. This allows the now reduced Trx to leave with reduced dithiols in its active site 211. This mechanism is made possible because of the selenocysteine residue. Selenocysteine is a cysteine analog that replaces sulfur with selenium. This maintains many chemical properties of the amino acid, however the pka of the selenol group in selenocysteine is significantly lower than the thiol group in cysteine (cysteine pka = ~5.8 vs. selenocysteine pka = ~8.3). This property allows the selenol to exist primarily as the ionized selenolate at physiological conditions, whereas the thiol exists in the protonated form 213. The ionized molecule is more reactive than the protonated version. In fact, replacement of selenocysteine with cysteine dramatically decreases activity of TrxR Functions of the Thioredoxin System As one of the major thiol-dependent electron donor systems in the cell, the Trx system carries out multiple important functions. The Trx system can perform mechanisms of both direct and indirect ROS scavenging. First, Trx can directly neutralize ROS. This action involves quenching hydroxyl radical by providing electrons, and requires Trx s catalytic cysteines, as shown by loss of ROS scavenging following mutation of these cysteines to alanine 215. Second, TrxR can indirectly scavenge ROS by providing electrons to other endogenous antioxidants, including ubiquinone. The reduced form of ubiquinone, ubiquinol, can act as a ROS scavenger by reducing radicals to protect from lipid peroxidation 216. The reduction of ubiquinone is catalyzed by TrxR under physiologic conditions, and may contribute to protection from oxidative stress 217. A critical function of the Trx system involves the regulation of protein disulfide formation. During oxidative stress ROS enact damage to cellular macromolecules. As proteins are the most highly concentrated macromolecule in cells, they are likely the most significant target of ROS 218. Not all

34 24 proteins are equally susceptible to oxidative damage, as subcellular localization and content of oxidation-sensitive amino acids may contribute 219. The most oxidation-sensitive amino acid residue is cysteine, that readily forms disulfide bridges with other cysteine residues upon oxidation 220. These aberrant disulfides can cause structural changes that impact protein function. Trx maintains protein redox status via its reductase activity which can reduce disulfides in an extensive number of target proteins to dithiols 221. Trx targets regulate a vast array of cellular processes including proliferation, metabolism, and apoptosis. In general, the role of Trx is to provide a reducing environment to protect from oxidative stress to prevent cell death and promote cell growth. The most important targets of Trx are reductive enzymes including Prx, ribonucleotide reductase, and methionine sulfoxide reductase. This highlights the importance of Trx in the maintenance of other reductive pathways 222. Prxs are a peroxidase family that act primarily as cellular antioxidants. Three subgroups of Prx exist: 2-cysteine Prx isoforms (Prx1-4), atypical 2-cysteine Prx isoform (Prx5), and 1-cysteine Prx isoform (Prx6) 223. Reduced Trx is the electron donor for Prx1-5 which can then remove ROS such as hydrogen peroxide, lipid hydroperoxides, and peroxynitrite. Hyperoxidation of Prx occurs from ROS overexposure. The catalytic cysteine residues oxidize to form disulfide bonds, which must be reversed by Trx 224. This establishes a redox cascade where the flow of electrons travels from NADPH through the Trx system to Prx and finally to neutralize toxic ROS 225. The speed of hydrogen peroxide scavenging by Prx is comparable to other significant ROS scavenging enzymes like glutathione peroxidase, on the magnitude of M -1 s -1, indicating that it is a major contributor to peroxide scavenging in cells 226,227. Prxs can inhibit cancer development or promote tumorigenesis depending on the specific Prx member and cancer context 228. However, high levels of Prxs are often associated with radioresistance and chemoresistance. For example, high Prx2 levels correlates to radioresistance in breast cancer cells, as well as with cisplatin resistance in

35 25 gastric cancer cells 229,230. Silencing of Prx2 sensitizes colorectal cancer cells to both radiation and oxaliplatin treatment 231. These data indicate that Trx may have an indirect role in promoting therapy resistance through Prx. Another significant reductase target of Trx is ribonucleotide reductase (RNR). RNR reduces all four ribonucleotides to deoxyribonucleotides for use in DNA synthesis. During ribonucleotide reduction the two active site cysteines of RNR form a double bond which must be reduced by a C- terminal dithiol through a thiol-disulfide exchange reaction 232. The original conformation of the C-terminal cysteines is restored by an external reductase system, either Trx or Grx which both have similar catalytic efficiency with mammalian RNR This redundancy allows for cells to withstand the loss of one reductase system. This is supported by studies showing that TrxR downregulation causes no change in deoxyribonucleotide triphosphate pools in mouse cancer cells 236. Some differences may exist as Trx is believed to support S phase electron donation, whereas Grx is believed to support DNA repair and mitochondrial DNA synthesis 235. Interestingly, Grx is an important effector of the GSH system, supporting compensatory mechanisms between Trx and GSH. Methionine is another oxidation-sensitive amino acid. However, protein reduction by Trx is limited to disulfides. Methionine sulfoxide reductase (Msr) is capable of reducing oxidized methionine in proteins and acquires its reducing potential from Trx 237. Although there are multiple Msr isoforms with different numbers of catalytic cysteines and mechanisms, they all require Trx to maintain their reductive capacity 222. This is yet another pathway by which Trx is required to protect cells from damage caused by oxidative stress. Beyond reductase systems, Trx is responsible for post-translational modifications to regulate numerous signaling cascades and transcription factors. Apoptosis signaling kinase 1 (ASK1) is

36 26 part of the MAPKKK family that activates the JNK and p38 MAP kinase pathway 238. Reduced Trx can bind ASK1 to inhibit its activation, directing it to ubiquitin-mediated degradation, which prevents ASK1 from activating proapoptotic gene expression 239. Inactivation of PTEN, a negative regulator of the PI3K-Akt pathway, results in the stimulation of proliferation. Trx1 binds PTEN and inhibits its activity, which may stimulate cancer cell proliferation 240. These data indicate a potential role for Trx in cancer proliferation and highlight the importance of Trx inhibition in cancer therapy Targeting the Thioredoxin System in Cancer A novel strategy for pancreatic cancer therapy may involve Trx system inhibition, as the Trx system is involved in protection from oxidative stress that may promote cancer. Excluding downstream targets of Trx, three major targets can be identified within the Trx system: NADPH, Trx, and TrxR. Since NADPH is ubiquitous and required in a vast array of major biosynthetic pathways, complete inhibition of NADPH production would not be feasible. The individual pathways of NADPH production could potentially be targeted. The molecule dichloroacetate inhibits the oxidative phase of the PPP and has been clinically tested 241,242. However, this likely may not be effective in a KRAS driven tumour like pancreatic cancer since the oxidative phase of the PPP is less involved in NADPH production than other pathways 108. When targeting the serine biosynthesis pathway PHGDH knockdown in overexpressing cancers inhibits proliferation 111. However, inhibitors of PHGDH, including PKUMDL-WQ-2101 and PKUMDL-WQ-2201, have only recently been identified and require further development before use in humans 243. Targeting glutamine metabolism through inhibition of ME1 has shown to be promising by inducing senescence and apoptosis in vitro 244. However, an inhibitor of ME1 has yet to be characterized. An inhibitor of the mitochondrial isoform ME2, NPD389, has been described but not developed 245.

37 27 As the effector of most functions of the Trx system, Trx would seem to be an ideal target. The synthetic molecule PX-12 is the most developed Trx inhibitor that covalently binds to Trx cysteine residues, forming mixed disulfides. PX-12 inhibits breast cancer cell proliferation in vitro and reduced tumour volume in vivo. Interestingly, PX-12 can be a substrate for TrxR causing competitive inhibition of TrxR for its normal substrate, oxidized Trx 246. PX-12 has been tested in clinical trials, and was shown to decrease plasma Trx in a phase I trial 247. However, a phase II trial of PX-12 in patients with advanced pancreatic cancer refractory to gemcitabine therapy showed no antitumour activity 248. These results may stem from the inability to achieve adequate plasma concentrations of PX-12 without significant toxicities because of rapid binding to plasma components. Combined results of clinical trials suggest that development of PX-12 for intravenous infusion is not feasible 249. The third and last potential target, TrxR, may contain the most potential for anticancer therapy. High TrxR1 expression correlates to poor prognosis in hepatocellular carcinoma, breast, and pancreatic cancer, among others 250,251. TrxR1 inhibition following sirna knockdown dramatically reduced tumour progression in xenograft models of lung cancer 252. In contrast, normal adult tissues can survive TrxR1 loss 253. Recent development has centered around attempts to develop specific inhibitors of TrxR1 with some success 254. However, these molecules require extensive development and testing before potential clinical use. The pan-trxr inhibitor auranofin is an FDA approved drug for the treatment of rheumatoid arthritis that has recently garnered interest for repurposing in cancer therapy 255. Auranofin is an orally available gold complex that irreversibly inhibits all TrxR isoforms, and is inexpensive, well-characterized, and has relatively minor toxicities 256. Although it has been reported to inhibit proteasome deubiquitinases in addition to TrxR 257, evidence to support this is sparse and seems to be secondary to cell death. Auranofin is capable of inducing apoptotic cell death following the proliferation of ROS 258,259. However, it

38 28 appears to work best in combination with other drugs after being testing as a combination therapy against various cancers In addition, auranofin has been used in phase I and II clinical trials for chronic lymphocytic leukemia therapy (see NCT ). Therefore, auranofin is likely the most ideal candidate to develop for pancreatic cancer therapy that targets the Trx system to augment oxidative stress. 1.4 A New Therapeutic Strategy for Pancreatic Cancer The hallmarks of cancer describe essential alterations involved in tumour development. They include the sustainment of proliferative signaling, evasion of growth suppressors, activation of invasion and metastasis, enabling replicative immortality, induction of angiogenesis, resisting cell death, deregulation of cellular energetics, and avoidance of immune destruction 263. Redox signaling and the responses to oxidative stress are involved in every hallmark of cancer 264. Therefore, consideration for cancer promotion by antioxidants should be made when developing novel anti-cancer therapies. Normal tissues are resistant to antioxidant loss, which may prevent harmful toxicities of antioxidant inhibition. As previously discussed, BSO and auranofin appear to be useful drug candidates for combined targeting of the GSH and Trx systems. Evidence suggests that these systems overlap to compensate and protect against inefficiencies or inhibition of either system. The Trx system can be an alternative route for cells to reduce oxidized GSH in vivo 265. In addition, physiologic concentrations of GSH and Grx are capable of reducing oxidized Trx in vitro, suggesting GSH as a backup to TrxR function 266. However, this observation was made in a cell-free system which may limit its applicability in the complex cell environment.

39 29 Combination therapy using both BSO and auranofin can help to overcome compensation. Ideally the combination of the two drugs will produce improved effectiveness compared to monotherapy with either compound. This result can either be stated as synergy or potentiation. Synergy is observed when the effect of two or more drugs is greater than the sum of effects when the drugs are administered as monotherapies. The effect is mutual in that each drug enhances the effect of the other 267. In contrast, a scenario by which the enhancement of effect is one-sided can be deemed a potentiation or augmentation. Evaluation of a potentiation can be stated as a fold change in the effect of one drug, caused by the effect of the other drug 267. Previous studies have shown that dual-inhibition of the GSH and Trx systems can lead to greater anti-cancer responses 127. Use of BSO and auranofin in combination sensitized mouse tumours to radiotherapy 83, and mesothelioma cells show extensive oxidative stress and cell death following combination of the drugs 268. These studies do not evaluate potential synergism of the combination of BSO and auranofin. Nor do any studies evaluate this combination in pancreatic cancer, which seems to be an exceptional target of pro-oxidant therapy. Considering that the Trx and GSH systems comprise the main antioxidant systems in the cell, I hypothesize that combination therapy using auranofin and BSO will cause primary pancreatic cancer cells to undergo apoptosis by exacerbating oxidative stress. Auranofin will be identifiable in pancreatic PDX tumours using IMC. I have three aims to test this hypothesis: Aim 1: Evaluate the effect of combination therapy on cell viability and death. Aim 2: Evaluate the effect of combination therapy on the redox environment of the cell. Aim 3: Evaluate the drug-target interaction of auranofin-trxr in vitro and biodistribution of auranofin in vivo.

40 Chapter 2 Targeting the Thioredoxin and Glutathione Antioxidant Systems to Treat Pancreatic Cancer Francesco Fazzari 1, Samir Barghout 2, Qing Chang 3, David Hedley 1,2,4 1 Department of Laboratory Medicine and Pathobiology, University of Toronto, Toronto, Ontario, Canada 2 Department of Medical Biophysics, University of Toronto, Toronto, Ontario, Canada 3 Fluidigm Canada Inc, Markham, Ontario, Canada 4 Division of Medical Oncology and Hematology, Princess Margaret Cancer Centre, Toronto, Ontario, Canada The work presented in this chapter is presented as a manuscript for future submission. 30

41 Introduction Pancreatic cancer is the 4 th leading cause of cancer death in Canada, however it is only the 12 th most commonly diagnosed 2. There is poor prognosis with an overall 5-year survival rate of 5% 4. Significant advancements in the development of new treatment options for the disease have not been made. Many patients with advanced disease are given standard chemotherapy drugs that provide little benefit, leading to an urgent need for new strategies to combat this disease. Pancreatic cancer, among other cancers, is characterized by excessive oxidative stress 48, which occurs when production of intracellular reactive oxygen species (ROS) outpaces the capacity of antioxidants to protect from ROS. These ROS are produced in mitochondria at an extreme rate in transformed cells due to inherently high metabolism, and may promote aggressive tumour features 50. Adaptations to protect from ROS are required for cancer cells to survive. These adaptations primarily involve oxidative stress signaling through Nrf2, which mediates the oxidative stress response to promote the production of the antioxidants glutathione (GSH) and thioredoxin (Trx) 54. The main oncogenic driver of pancreatic cancer, KRAS, induces Nrf2 activation to maintain antioxidant defenses 47. In conjunction to stimulating antioxidant production, pancreatic cancer relies on glutamine metabolism for the production of NADPH 108. This adaptation occurs because of the need to protect against high ROS during oxidative stress, as NADPH provides reducing power to antioxidant systems in the cell to prevent cytotoxic damage caused by high ROS levels. ROS exhibit dual roles; lower concentrations of ROS that are promoted by antioxidants can stimulate cancer cell proliferation, whereas higher concentrations of ROS are damaging to cells 49,52. Therefore, antioxidants may contribute to aggressive growth and metastasis in cancer because of their functions in ROS scavenging, preventing ROS-mediated cellular damage, and

42 32 promoting ROS-mediated proliferative signalling 148,269. The reliance of pancreatic cancer on antioxidants to protect from oxidative stress may provide a unique vulnerability in the tumour. Targeting antioxidants may promote cytotoxic damage specifically in tumours by augmenting oxidative stress. Both the GSH and Trx systems have been extensively studied and tested inhibitors exist. GSH is the most abundant antioxidant cofactor in cells and is responsible for ROS scavenging 148. The ratelimiting step of GSH synthesis is catalyzed by glutamate-cysteine ligase (GCL) 148. An inhibitor of GCL, buthionine sulfoximine (BSO), has been well characterized and previously tested in human subjects with minimal toxicity 197,198. Trx reduces aberrant protein disulfides caused by oxidative stress 269. Oxidized Trx is recycled by thioredoxin reductase (TrxR) using NADPH to replenish its reducing power 256. Three TrxR isozymes exist: cytosolic TrxR1 that has been identified as a fitness gene 209, mitochondrial TrxR2, and testes-specific TrxR3. The gold complex auranofin is an inhibitor of all three TrxR proteins and has previously been used in the treatment of rheumatoid arthritis 255. It has garnered interest for repurposing as a therapy in various cancers 260,261,270. Many recent studies have identified oxidative stress as an effective target in cancer therapy, including in KRAS driven tumours 75,128, Targeting one antioxidant pathway may allow for compensation from another to continue providing protection to the cell 204. This may possibly be overcome by a double hit. To our knowledge no studies have attempted a double hit in pancreatic cancer. Considering that Trx and GSH comprise the main antioxidant systems in the cell, combination therapy using auranofin and BSO may cause cell death by depleting antioxidant capacity and increasing ROS in primary pancreatic cancer cells.

43 Materials & Methods Cell Culture and Reagents Primary cell lines (9A and OCIP23) were derived from human pancreatic cancer tumours. 9A was maintained in Roswell Park Memorial Institute (RPMI) 1640 (Wisent) medium supplemented with 10% fetal bovine serum (FBS) (VWR Seradigm Life Science) and 1x penicillin/streptomycin (Wisent). OCIP23 was maintained in Dulbecco s Modified Eagle Medium (DMEM)/F-12 (Wisent) supplemented with 2.5% FBS and 1x penicillin/streptomycin. Cells were cultured at 37 C in a humidified incubator with 5% CO2. Auranofin (Sigma-Aldrich) was diluted in DMSO (Sigma-Aldrich) at 100mM, aliquoted and stored at -10 C to be used once without refreezing. BSO (Sigma-Aldrich) was prepared fresh in sterilized MilliQ water at 100mM. MTS Assay Cell viability was assayed by MTS assay and performed according to the manufacturer s protocol (Promega). In summary, 96 well plates were seeded with 5000 cells/100 L/well and incubated for 24 hours. Medium was replaced with medium containing DMSO (Sigma-Aldrich) [0.003% for OCIP23, 0.004% for 9A], auranofin, BSO, or combination of auranofin and BSO at the indicated concentrations. Each treatment was performed in triplicate. Cells were incubated for 24 hours, then cell viability was determined by adding 20 L of MTS reagent followed by incubation for 3 hours at 37 C and absorbance reading at 492nm using the Multiskan EX microplate spectrophotometer (Thermo Fisher). Clonogenic Assay

44 34 Clonogenic assays evaluated drug treatment effects on colony formation and was performed as previously described 274. Briefly, cells were grown to 70-80% confluence in 10cm plates then treated with vehicle (1 L/mL DMSO), auranofin, BSO, or a combination at the indicated concentrations for 24 hours. Cells were harvested and reseeded in 6-well plates at 10 5, 10 4, 10 3, and 10 2 cells/well in triplicate, then allowed to incubate for 12 days before fixing and staining with 0.2% methylene blue in 50% ethanol/water. Colonies were counted using the Gel Count (Oxford Optronix) counter. Plating efficiency was calculated as the average number of colonies divided by the number of cells plated. Surviving fraction was expressed as the plating efficiency of the sample divided by the plating efficiency on vehicle control plates. Multiparametric Flow Cytometry Apoptosis, ROS levels, and GSH levels were assayed by flow cytometry. 7x10 5 cells/plate were seeded in 6cm plate. After 24 (OCIP23) or 48 hours (9A), media was replaced with either normal medium or BSO-supplemented medium. Auranofin was diluted in DMSO to 1000x the indicated concentrations and added at 1 L/mL. After 24 hour treatment the cells were trypsinized and resuspended in medium. Cells were centrifuged at 1100rpm for 5 minutes, then the pellets were resuspended in 0.5mL serum free media. Cell suspensions were then incubated in a water bath at 37 C with 40nM DiIC1(5) (ThermoFisher Scientific) and 5 M 2,7 -dichlorofluorescein (DCF) (ThermoFisher Scientific) for 30 minutes. 5 minutes before the end of the incubation, 40 M monobromobimane (mbbr) (ThermoFisher Scientific) and 1 g/ml propidium iodide (PI) (ThermoFisher Scientific) were added. Following incubation, samples were cooled on ice for 10 minutes then analyzed with the CytoFLEX flow cytometer (Beckman Coulter). During the time course experiment samples were prepared in the same manner, however the drug treatments were

45 35 for 2 and 5 hours, and mbbr was excluded from the panel of fluorescent probes. Data analysis was performed using FlowJo Software (BD). Tumour sample preparation Animal experiments were carried out according to protocols approved by the University Health Network Animal Care Committee. Patient-derived xenograft (PDX) mice were established subcutaneously in 5 week old severe combined immunodeficiency (SCID) mice. Two models, OCIP83 and 9A were used. Mice were given an intraperitoneal injection of 10mg/kg auranofin for a 24 hour treatment. Mice were later given 120mg/kg pimonidazole (Hypoxyprobe, Inc.), followed by 30mg/kg EF5 (obtained from Dr. Cameron J. Koch, University of Pennsylvania) 5 and 3 hours prior to sacrifice, respectively. Tumours were excised, fixed, and processed for paraffin embedding. Immunohistochemistry (IHC) Formalin-fixed, paraffin-embedded (FFPE) sections of subcutaneous PDX tumours were incubated with mouse pimonidazole antibody (Hypoxyprobe, Inc.) at 1:400, overnight at 4 C. Biotinylated horse anti-mouse IgG was added at 1:200 concentration and incubated for 1 hour. Detection was performed using the HRP labeling (Vector Labs) and DAB chromogen (Dako). Imaging Mass Cytometry (IMC) Staining, Data Acquisition, and Visualization Imaging mass cytometry was used to directly image the distribution of atomic gold in tissue sections following treatment with auranofin. Tissue staining for IMC was performed as previously described 275. In summary, FFPE sections were dewaxed and rehydrated as routine. Antigen retrieval and blocking with BSA was performed, followed by incubation with primary antibody

46 36 against pimonidazole for 1 hour at room temperature (RT). Samples were then incubated overnight at 4 C with a metal-conjugated antibody cocktail diluted in PBS/0.5% BSA (Table 1). Following this, samples were washed and exposed to 25 M Ir-Intercalator (Fluidigm) for 30 minutes at RT. Lastly, the samples were rinsed then dried before IMC ablation. Regions of interest were determined by assessing IHC samples stained for pimonidazole. During data acquisition, the IMC immunostained samples are inserted into a laser ablation chamber where the tissue is ablated in a 1 m diameter spot. The ablation spot is vaporized, and the plume is carried with high time-fidelity into the inductively coupled plasma ion source for simultaneous analysis by the mass cytometer (Fluidigm). The metal isotopes located on each spot are simultaneously measured. This process continues for the entire area, eventually yielding an intensity map of the targets throughout the tissue region of interest. Data-processing and image visualization was performed using the in-house-developed MCD Viewer (Fluidigm). The signal data from the mass cytometer was exported in text format, then reconstructed into images. The epithelial cells were defined by E-cadherin and pan-keratin staining, and the stromal cells were defined by collagen and SMA. Areas within these regions were selected in the MCD viewer to semi-quantitatively determine the average number of gold ions within them Cell Lysis and Protein Determination Lysis was performed using a lab-made lysis buffer containing 50mM Hepes (ph 8.00), 10% glycerol, 1% Triton-X, 150mM NaCl, 1mM EDTA, 1.5mM MgCl2, 100mM NaF, 10mM Na4P2O7 10H2O, 1mM Na3VO4, and 1 tablet/7ml buffer complete protease inhibitor cocktail (Sigma-Aldrich). Cell cultures were incubated with 700 L lysis buffer for 1 hour at 4 C, then

47 37 collected and centrifuged at 14,000rpm for 15 minutes at 4 C. Supernatants were collected and stored at -80 C until use. Protein determination was performed using the BCA Assay kit (ThermoFisher Scientific) as per the manufacturer s protocol. Gel Electrophoresis and Immunoblot Samples were resolved using 10% SDS-PAGE. Following transfer to polyvinylidene difluoride membranes, proteins were detected with antibodies to TrxR1 1:1000 (Cell Signaling Technology), TrxR2 1:1000 (Cell Signaling Technology), -actin 1:7000 (Abcam), GAPDH 1:4000 (Cell Signaling Technology). Horseradish peroxidase (HRP) conjugated secondary sheep anti-mouse (Abcam) and donkey anti-rabbit (Abcam) were used for detection at 1:5000. Quantification was performed by densitometry using ImageJ (NIH). Cellular Thermal Shift Assay Cellular Thermal Shift Assay (CETSA) was performed as previously described 276 to assess auranofin-trxr engagement. Briefly, to determine the optimal temperature at which maximal thermal shifts were observed, cells were treated with vehicle or 10 M auranofin for 1 h, then divided into equal portions, centrifuged, and resuspended in PBS with protease inhibitors. Cells were then heated at temperatures ranging from C for 3 min in a thermal cycler (SimpliAmp, Applied Biosystems), cooled at RT for 3 min, snap-frozen in liquid nitrogen, thawed at RT, and then lysed by 3 freeze/thaw cycles with vortexing in between. Samples were then centrifuged at 20,000 x g for 20 min at 4 C and supernatants were collected. Lysates were analyzed by immunoblotting. When evaluating changes with different auranofin concentrations the same procedure was followed, however cells were treated with increasing auranofin concentrations at the determined optimal temperature of 56 C.

48 38 Statistical Analysis Data are presented as the mean value S.E.M. GraphPad Prism version 7.0 (GraphPad Software, Inc., San Diego, CA, USA) was used for statistical analysis. A value of P<0.05 was considered to indicate statistically significant differences. 2.3 Results Auranofin, but not BSO, inhibits cell growth and viability of primary pancreatic cancer cell lines. To evaluate the effect of Auranofin and BSO on cell viability, human primary pancreatic cancer cells were treated with different concentrations of either drug for 24 hours. MTS assay measured cell metabolic activity. BSO alone displayed no significant effect on the viability of the primary cells (Fig. 2.1A). In contrast, auranofin reduced viability in a concentration dependent manner with IC50 values 1.40μM and 2.35μM in OCIP23 and 9A, respectively (Table 2.1; Fig. 2.1B). These data suggest that after 24 hours, these two primary pancreatic cancer cell lines are susceptible to auranofin, but not BSO. Since BSO does not reduce viability at this time point, any interaction between the two drugs cannot be deemed synergy. BSO potentiates auranofin s inhibitory effect on cell viability of primary pancreatic cancer cells. Using the MTS assay, we evaluated cell viability following treatment of cells with a combination of auranofin and BSO. Co-treatment with 10μM BSO caused a 61.1-fold decrease in the IC50 of auranofin in OCIP23 (IC50 = 22.9nM), and a 4.07-fold decrease in 9A (IC50 = 578nM) (Fig. 2.1B). The addition of 50μM BSO caused a 238-fold decrease in the IC50 of auranofin in OCIP23 (IC50 = 5.88nM), and a 11.0-fold decrease in the IC50 in 9A (IC50 = 213nM) (Fig. 2.1B). These

49 39 data support a strong BSO-mediated potentiation of auranofin cytotoxicity. The higher IC50 and lower potentiation observed with 9A cells suggest they possess a higher reserve capacity of TrxR1/2 activity that makes them more resistant to drug-induced oxidative stress. To test this hypothesis, we assessed expression levels of Trx1/2 in 9A and OCIP23 cells. 9A cells showed a significantly higher expression level of TrxR1, but not TrxR2, compared to OCIP23 cells (p=0.0007) (Fig. 2.1C and D). This may explain, at least in part, the differential response of OCIP23 and 9A to auranofin and combination with BSO. Combination of BSO and auranofin inhibits clonogenicity of primary pancreatic cancer cells. We then conducted clonogenic assay to determine whether the BSO-auranofin combination inhibited long-term viability and if low auranofin concentrations on their own improved viability as observed by MTS assay. There was a significant decrease in clonogenicity following combination treatment in 9A (p=0.0155) with 43.1% surviving fraction (Fig. 2.1E) and OCIP23 (p=0.0088) with 43.8% surviving fraction (Fig. 2.1F) when compared to vehicle controls. Single treatment with auranofin (100nM or 400nM) showed no increase in clonogenicity, which contrasts the improved viability of low auranofin concentrations observed via MTS assay (Fig. 2.1B). The results of these two assays may differ because of the different endpoints measured, metabolism versus colony growth. These data support the potentiation of auranofin by BSO at low auranofin concentrations that otherwise have no effect on cell death and viability when used as monotherapy. Table 2.1. IC50 values of auranofin monotherapy and in combination with BSO in primary pancreatic cancer cell lines. Cell Line BSO Concentration ( M) Auranofin IC50 (nm) OCIP OCIP OCIP A A A

50 Surviving Fraction (%) % Cell Viability % Cell Viability 40 A) B) C) BSO Dose-Response OCIP23 9A BSO concentration (um) Auranofin Dose-Response with and without BSO OCIP23-0 BSO OCIP23-10uM BSO OCIP23-50uM BSO 9A - 0 BSO 9A - 10uM BSO 9A - 50uM BSO Auranofin concentration (nm) D) TrxR1 β-actin TrxR2 β-actin 9A OCIP23 E) F) 140 OCIP ** Vehicle BSO 10uM Au 400nM Au 100nM Combination Figure 2.1. Dose-response of antioxidant pathway inhibitors, TrxR isozyme levels, and colony formation potential. (A and B) Viability of primary pancreatic cancer cells was determined after 24 hour treatment using an MTS assay. Shown are the toxicity profiles of A) BSO monotherapy (n=3), and B) auranofin monotherapy or combination (n=3). (C and D) Relative amounts of C) TrxR1 and TrxR2 in untreated cell lines (n=3) and their corresponding D) immunoblots. Levels were normalized to the first OCIP23 sample. (E and F) Colony formation of primary cell lines was determined after 24 hour treatment. Cells were reseeded and incubated for 12 days before colonies were assesed for E) 9A (n=3) (Combination of Au 400nM and BSO 10 M) and F) OCIP23 (n=3) (Combination of Au 100nM and BSO 10 M). Combinations were compared to vehicle controls. Surviving fraction is represented as a percentage of vehicle controls. Shown are the S.E.M. Significant decrease from vehicle control, * = p<0.05, ** = p<0.01, *** = p<0.001.

51 41 BSO-auranofin combination induces apoptosis in primary pancreatic cancer cells. We conducted live cell flow cytometry to determine whether apoptosis contributed to cell death after BSO-auranofin combination. The loss of mitochondrial membrane potential (MMP) has been observed to be one of the first steps of apoptosis, and has previously been observed following auranofin treatment 258. In addition, the loss of membrane integrity, and subsequent uptake of propidium iodide (PI), is a marker of cell death. Here we used DiIC1(5), a marker of MMP, and PI to assess cell death after combination treatment. Auranofin or BSO alone did not increase the MMP negative or PI positive populations. In contrast, combinations of these low concentrations increased the proportion of cells observed to be starting apoptosis and those that have already died in both OCIP23 (Fig. 2.2) and in 9A. These results indicate that combinations of low concentrations of auranofin and BSO induces apoptosis in primary pancreatic cancer cells. BSO-auranofin combination reduces thiol levels and induces an acute increase in ROS levels. We also used live cell flow cytometry to assess the effect of combinatorial treatment on oxidative stress. First, GSH was examined after 24 h treatment by using mbbr, which fluoresces after binding to free thiols. GSH is the most prominent free thiol in cells, and contributes to approximately 50% of the mbbr signal 277. The treated cells were gated for PI negative populations to restrict analysis to viable cells. Auranofin did not affect GSH levels when applied as a monotherapy in OCIP23 but had a significant impact in 9A (p=0.0345). However, BSO had a significant impact (OCIP23 p<0.0001; 9A p=0.0003) on GSH levels (Fig. 2.4A). The GSH status of MMP negative populations, both PI negative and positive, was assessed during the same live cell flow cytometry assay. Following 24 hour treatment both populations displayed depleted GSH compared to MMP positive populations (Fig. 2.3). These data suggest that the combination therapy may lead to cytotoxicity by depletion of GSH levels.

52 PI 42 Auranofin concentration 0nM 6nM 30nM 75nM BSO concentration 0μM 10μM 50μM MMP Figure 2.2. Representative two parameter plots measuring cell death and apoptosis in OCIP23. Cells were treated with either monotherapy or combination therapy for 24 hours followed by flow cytometry. Apoptotic cell populations are MMP-, PI- (bottom left quadrant); Dead cell populations are MMP-, PI+ (Upper left quadrant); Live cells are MMP+, PI- (bottom right quadrant). Shown are the percentages of the total population for each subpopulation.

53 MMP 43 Auranofin concentration 0nM 6nM 30nM 75nM BSO concentration 0μM 10μM 50μM mbbr Figure 2.3. Representative two parameter plots assessing the GSH content of apoptotic cells in OCIP23. Cells were treated with either monotherapy or combination therapy for 24 hours followed by flow cytometry. MMP is a marker for apoptosis, mbbr is a marker for GSH, and PI is a marker for cell viability. PI positive (non-viable) cells are shown in red, and PI negative (viable) cells are shown in grey. Second, ROS were examined following 2 and 5 hour treatments by 2,7 -dichlorofluorescin (DCF), which fluoresces following oxidation by ROS. Again, treated cells were gated for PI negative populations to restrict analysis to viable cells. At these time points, ROS were not affected by auranofin alone. However, BSO alone and in combination with auranofin displayed a significant dose-dependent increase (OCIP23 p<0.0001; 9A p<0.0001) in ROS levels (Fig. 2.4B). Combined, these results indicate that the combination of auranofin and BSO leads to oxidative stress in the

54 44 treated cells. This suggests that depletion of GSH and increase in ROS may contribute to cell death after combination therapy. A) B) Figure 2.4. Effect of treatment on thiol and ROS levels in 9A and OCIP23 cell lines. A) Mean fluorescence intensity (MFI) of mbbr in PI negative cells after 24 hour treatment with BSO, auranofin, or combination (9A n=3; OCIP23 n=3). B) MFI of DCF in PI negative cells after 2 and 5 hour treatment with BSO, auranofin, or combination (9A n=3; OCIP23 n=3). Fluorescence was normalized to non-treated controls. Shown are the S.E.M. Auranofin binds to TrxR1 in primary pancreatic cancer cells at concentrations associated with oxidative stress and cell death. We performed a cellular thermal shift assay (CETSA) to observe whether auranofin engages its targets TrxR1 and TrxR2 in intact primary pancreatic cancer cells at different concentrations (Fig.

55 45 2.5). This assay is based on thermostabilization of protein targets upon binding chemical ligands 276. It is more clinically relevant as it allows to test drug-target engagement in the complex cellular environment, as opposed to cell-free assays. As an initial step in CETSA, we treated 9A cells with a saturating concentration (10 µm) of auranofin and determined the optimal temperature at which we observed maximal thermostabilization of TrxR1/2 in treated cells compared to untreated controls. By analysis of thermal shift blots, we found 56 C to be the optimal temperature for TrxR1 and TrxR2 (Fig. 2.5A). We then tested auranofin engagement after treating 9A and OCIP23 cells with increasing concentrations to determine whether it binds TrxR1/2 at concentrations associated with cell death. In both 9A and OCIP23 we observed TrxR1-auranofin binding at concentrations as low as 78nM (Fig. 2.5B and C). Binding was observed at concentrations much below the IC50 s of auranofin monotherapy in both cell lines. In addition, our CETSA data suggest that auranofin-mediated cell death is attributable to targeting TrxR1, but not TrxR2, at IC50 s observed after BSO combination (Table 2.1). Since OCIP23 and 9A cells exhibit different TrxR1 but not TrxR2 expression levels, our CETSA data provide further explanation of greater potentiation observed with OCIP23 compared to 9A cells. Taken together, our data show that auranofin bound TrxR1 in intact primary pancreatic cancer cells at concentrations capable of inducing cell death, and auranofin targets TrxR1 not TrxR2 at concentrations associated with BSO-mediated potentiation.

56 46 A) 46 C UT T 49 C UT T 52 C UT T 55 C UT T 58 C UT T 61 C UT T 64 C UT T 67 C UT T 70 C UT T TXRX1 TXRX2 GAPDH B) Auranofin (µm) TRXR1 TRXR2 GAPDH C) Auranofin (µm) TRXR1 TRXR2 GAPDH Figure 2.5. Auranofin-TrxR interaction in vitro. A) Engagement of auranofin and TrxR was assessed by CETSA. 9A was treated with 10 M auranofin for 1 hour. Following drug treatment, cells were exposed to the indicated temperatures for 3 minutes prior to lysis and immunoblot (n=1). B) 9A and C) OCIP23 were treated with different auranofin concentrations for 2 hours before heat treatment at 56 C and immunoblot (n=1).

57 47 Table 2.2. Panel of markers measured by IMC using subcutaneous PDX tumours. Antibody/Reagent Antibody Clone Metal Final Concentration ( g/ml) Avanti Lipid N/A 115 In 10 M -SMA 1A4 141 Pr 1.25 Thioredoxin 2G11/TRX 146 Nd 10 EF5 ELK Sm 10 Pimonidazole N/A 155 Gd 10 E-cadherin 24E Gd 2.5 Pan-keratin C Dy 0.6 Goat Anti-Mouse N/A 165 Ho 1:100 (Erg1 detection) Collagen Goat polyclonal 169 Tm 1.25 Intercalator N/A 191/193 Ir 0.25 M Auranofin N/A 197 Au 10mg/kg Auranofin distributes within PDX tumours 24 hours after treatment, and associates with the stroma. Auranofin contains gold in its molecular structure, which suggested the possibility to detect it in tissues using imaging mass cytometry (IMC), as we have recently described for platinum-based drugs 275. This novel high throughput technology detects heavy metal isotopes on a histological section. These heavy metals can either be linked to antibodies or found within the tissue. Data gained from this method can give information on the distribution of auranofin, its interactions with the tumour tissue, and an approximation of its concentration within different compartments of a tumour as long as auranofin is detectable with IMC. The antibody panel used can be seen in Table 2.2. In this pilot study we observed auranofin distributed throughout the two PDX tumours (Fig. 2.6). Auranofin appeared to be more closely associated with collagen in the stromal compartment of the tumour (Fig. 2.6A and C), rather than E-cadherin in the epithelial compartment (Fig. 2.6B and D). We then approximated the average number of gold ions per pixel (each pixel has a defined volume of 5 m 3 ) in each compartment (Table 2.3). This semi-quantitative method showed a

58 9A OCIP83 48 preferential accumulation of auranofin in the stroma. This method can be expanded in the future to gain a greater understanding of auranofin s effects in vivo. A B 200μm 200μm C D 200μm 200μm Figure 2.6. Distribution of auranofin in subcutaneous PDX tumours. Mice were treated with 10mg/kg auranofin for 24 hours, followed by pimonidazole for 5hr and EF5 for 3 hours prior to sacrifice. Tissue section ablations were performed using IMC and markers were false-coloured. The relation of auranofin (green) to (A and C) stromal collagen (blue), (B and D) epithelial E- cadherin (blue), and pimonidazole (red) in OCIP83 (A and B) and 9A (C and D) tumours is shown. Scale bars are provided for reference. Arrows highlight regions of interest.